Articles

The why

• Why programming for the browser needs a different kind of language JavaScript wasn't built for today's browser. scrml is. Six things a browser-language should own at the type level.
• Components are states Why the UI of a scrml app should be a fully-handled state machine — and why that's a language-level claim, not a framework convention.
• The server boundary disappears What changes when the compiler partitions client and server automatically: no API routes, no fetch wrappers, no request/response types.

The what

• The Tier 0/1/2 promotion ladder From if= chains to <match> blocks to <engine> state machines — and the mechanical promotion CLI between them.
• Why scrml has to deprecate function and component overloading Overloading is library-syntactic-sugar; structural exhaustiveness is a language commitment. The two are in tension. The trio that replaces overloading.
• CSS without a build step Native @scope reached Baseline. A browser-language designed today should compile its scoped styles to that, not to a runtime mangler.
• Mutability contracts Who can write to this value? In mainstream stacks it's folklore. In scrml it's a compile error. Five layers of write-authority, all compile-time.
• Realtime & workers Socket.io vs. Pusher vs. Liveblocks for realtime; BullMQ vs. Inngest for queues; Comlink vs. raw Worker boilerplate. scrml replaces the whole matrix with two language primitives.

The why-not (existing tools)

• What npm package do you actually need in scrml? Worked through one ecosystem tier at a time: framework, routing, styling, validation, SQL, realtime — what scrml subsumes, what's left.
• The ORM trap Why an ORM trades one kind of SQL noise for another. ?{ } + compile-time schema introspection is the alternative.
• What scrml's LSP can do that no other LSP can SQL column completion. Cross-file go-to-def. Code actions for tier promotion. What vertical integration unlocks at the editor layer — and why giti follows from the same principle.

Release announcements

• scrml v0.3.0 — Approach A complete: per-route, per-role, content-addressed The big new thing: whole-stack closure analysis. Per-(entry-point, role, tier) chunks. <auth role> as a compile-time visibility constraint.
• scrml v0.2.0 — the foundational reactive surface V5-strict state, validators, the auto-synthesized validity surface, the pattern-matching ladder.
• Roadmap from v0.3 forward What's pending, what's in flight, what's next. Honest current-state framing.